`AttributeError: 'list' object has no attribute 'shape'` when using saliency map with teacher model and head distillation

[rezafuru]

Description:
When using a DataLoader that returns a tuple of (tensor, saliency_map) in __getitem__, and running head distillation with a teacher model (i.e., without bottleneck injection), the forward pass fails with the following error:

AttributeError: 'list' object has no attribute 'shape'

Cause:
The issue arises because the saliency-augmented input is not properly unpacked before being forwarded through the teacher model, which expects a plain tensor and not a tuple/list.

Traceback snippet:

File "patch_embed.py", line 50, in forward
  B, C, H, W = x.shape
AttributeError: 'list' object has no attribute 'shape'

Steps to Reproduce:

1. Use a dataset where __getitem__ returns (tensor, saliency_map)
2. Enable head distillation
3. Train using a teacher model (no bottleneck injection)

Expected Behavior:
The teacher model should correctly receive just the tensor (not the tuple) in the forward pass.